[PATCH 0/2] Two fixes for sysctl_sched_rr_timeslice

[PATCH 0/2] Two fixes for sysctl_sched_rr_timeslice

[Cyril Hrubis]

- Fixes rounding error for initial value with CONFIG_HZ_300

- Fixes read from the file after reset to default (by writing val <= 0)

Cyril Hrubis (2):
  sched/rt: Fix sysctl_sched_rr_timeslice intial value
  sched/rt: sysctl_sched_rr_timeslice show default timeslice after reset

 kernel/sched/rt.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

-- 
2.41.0

[PATCH 1/2] sched/rt: Fix sysctl_sched_rr_timeslice intial value

[Cyril Hrubis]

Thre is 10% rounding error in the intial value of the
sysctl_sched_rr_timeslice with  CONFIG_HZ_300=y.

This was found with LTP test sched_rr_get_interval01:

sched_rr_get_interval01.c:57: TPASS: sched_rr_get_interval() passed
sched_rr_get_interval01.c:64: TPASS: Time quantum 0s 99999990ns
sched_rr_get_interval01.c:72: TFAIL: /proc/sys/kernel/sched_rr_timeslice_ms != 100 got 90
sched_rr_get_interval01.c:57: TPASS: sched_rr_get_interval() passed
sched_rr_get_interval01.c:64: TPASS: Time quantum 0s 99999990ns
sched_rr_get_interval01.c:72: TFAIL: /proc/sys/kernel/sched_rr_timeslice_ms != 100 got 90

What this test does is to compare the return value from the
sched_rr_get_interval() and the sched_rr_timeslice_ms sysctl file and
fails if they do not match.

The prolem it found is the intial sysctl file value which was computed as:

static int sysctl_sched_rr_timeslice = (MSEC_PER_SEC / HZ) * RR_TIMESLICE;

which works fine as long as MSEC_PER_SEC is multiple of HZ, however it
introduces 10% rounding error for CONFIG_HZ_300:

(MSEC_PER_SEC / HZ) * (100 * HZ / 1000)

(1000 / 300) * (100 * 300 / 1000)

3 * 30 = 90

This can be easily fixed by reversing the order of the multiplication
and division. After this fix we get:

(MSEC_PER_SEC * (100 * HZ / 1000)) / HZ

(1000 * (100 * 300 / 1000)) / 300

(1000 * 30) / 300 = 100

Signed-off-by: Cyril Hrubis <chrubis@suse.cz>
CC: Jiri Bohac <jbohac@suse.cz>
---
 kernel/sched/rt.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/kernel/sched/rt.c b/kernel/sched/rt.c
index 00e0e5074115..185d3d749f6b 100644
--- a/kernel/sched/rt.c
+++ b/kernel/sched/rt.c
@@ -25,7 +25,7 @@ unsigned int sysctl_sched_rt_period = 1000000;
 int sysctl_sched_rt_runtime = 950000;
 
 #ifdef CONFIG_SYSCTL
-static int sysctl_sched_rr_timeslice = (MSEC_PER_SEC / HZ) * RR_TIMESLICE;
+static int sysctl_sched_rr_timeslice = (MSEC_PER_SEC * RR_TIMESLICE) / HZ;
 static int sched_rt_handler(struct ctl_table *table, int write, void *buffer,
 		size_t *lenp, loff_t *ppos);
 static int sched_rr_handler(struct ctl_table *table, int write, void *buffer,
-- 
2.41.0

[PATCH 2/2] sched/rt: sysctl_sched_rr_timeslice show default timeslice after reset

[Cyril Hrubis]

The sched_rr_timeslice can be reset to default by writing value that is
<= 0. However after reading from this file we always got the last value
written, which is not useful at all.

$ echo -1 > /proc/sys/kernel/sched_rr_timeslice_ms
$ cat /proc/sys/kernel/sched_rr_timeslice_ms
-1

Fix this by setting the variable that holds the sysctl file value to the
jiffies_to_msecs(RR_TIMESLICE) in case that <= 0 value was written.

Signed-off-by: Cyril Hrubis <chrubis@suse.cz>
CC: Jiri Bohac <jbohac@suse.cz>
---
 kernel/sched/rt.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/kernel/sched/rt.c b/kernel/sched/rt.c
index 185d3d749f6b..0597ba0f85ff 100644
--- a/kernel/sched/rt.c
+++ b/kernel/sched/rt.c
@@ -3062,6 +3062,9 @@ static int sched_rr_handler(struct ctl_table *table, int write, void *buffer,
 		sched_rr_timeslice =
 			sysctl_sched_rr_timeslice <= 0 ? RR_TIMESLICE :
 			msecs_to_jiffies(sysctl_sched_rr_timeslice);
+
+		if (sysctl_sched_rr_timeslice <= 0)
+			sysctl_sched_rr_timeslice = jiffies_to_msecs(RR_TIMESLICE);
 	}
 	mutex_unlock(&mutex);
 
-- 
2.41.0

Re: [LTP] [PATCH 1/2] sched/rt: Fix sysctl_sched_rr_timeslice intial value

[Petr Vorel]

Hi,

[ Cc Shile Zhang ]

> Thre is 10% rounding error in the intial value of the
> sysctl_sched_rr_timeslice with  CONFIG_HZ_300=y.

> This was found with LTP test sched_rr_get_interval01:

> sched_rr_get_interval01.c:57: TPASS: sched_rr_get_interval() passed
> sched_rr_get_interval01.c:64: TPASS: Time quantum 0s 99999990ns
> sched_rr_get_interval01.c:72: TFAIL: /proc/sys/kernel/sched_rr_timeslice_ms != 100 got 90
> sched_rr_get_interval01.c:57: TPASS: sched_rr_get_interval() passed
> sched_rr_get_interval01.c:64: TPASS: Time quantum 0s 99999990ns
> sched_rr_get_interval01.c:72: TFAIL: /proc/sys/kernel/sched_rr_timeslice_ms != 100 got 90

> What this test does is to compare the return value from the
> sched_rr_get_interval() and the sched_rr_timeslice_ms sysctl file and
> fails if they do not match.

> The prolem it found is the intial sysctl file value which was computed as:

> static int sysctl_sched_rr_timeslice = (MSEC_PER_SEC / HZ) * RR_TIMESLICE;

> which works fine as long as MSEC_PER_SEC is multiple of HZ, however it
> introduces 10% rounding error for CONFIG_HZ_300:

> (MSEC_PER_SEC / HZ) * (100 * HZ / 1000)

> (1000 / 300) * (100 * 300 / 1000)

> 3 * 30 = 90

> This can be easily fixed by reversing the order of the multiplication
> and division. After this fix we get:

> (MSEC_PER_SEC * (100 * HZ / 1000)) / HZ

> (1000 * (100 * 300 / 1000)) / 300

> (1000 * 30) / 300 = 100

> Signed-off-by: Cyril Hrubis <chrubis@suse.cz>
> CC: Jiri Bohac <jbohac@suse.cz>
> ---
>  kernel/sched/rt.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)

> diff --git a/kernel/sched/rt.c b/kernel/sched/rt.c
> index 00e0e5074115..185d3d749f6b 100644
> --- a/kernel/sched/rt.c
> +++ b/kernel/sched/rt.c
> @@ -25,7 +25,7 @@ unsigned int sysctl_sched_rt_period = 1000000;
>  int sysctl_sched_rt_runtime = 950000;

>  #ifdef CONFIG_SYSCTL
> -static int sysctl_sched_rr_timeslice = (MSEC_PER_SEC / HZ) * RR_TIMESLICE;
> +static int sysctl_sched_rr_timeslice = (MSEC_PER_SEC * RR_TIMESLICE) / HZ;

It looks like very old bug, from v4.11-rc1. I guess this should go to all stable
and LTS kernels.

Fixes: 975e155ed873 ("sched/rt: Show the 'sched_rr_timeslice' SCHED_RR timeslice tuning knob in milliseconds")

Reviewed-by: Petr Vorel <pvorel@suse.cz>

Kind regards,
Petr

>  static int sched_rt_handler(struct ctl_table *table, int write, void *buffer,
>  		size_t *lenp, loff_t *ppos);
>  static int sched_rr_handler(struct ctl_table *table, int write, void *buffer,

Re: [PATCH 1/2] sched/rt: Fix sysctl_sched_rr_timeslice intial value

[Mel Gorman]

On Wed, Jul 19, 2023 at 12:37:42PM +0200, Cyril Hrubis wrote:
> Thre is 10% rounding error in the intial value of the
> sysctl_sched_rr_timeslice with  CONFIG_HZ_300=y.
> 
> This was found with LTP test sched_rr_get_interval01:
> 
> sched_rr_get_interval01.c:57: TPASS: sched_rr_get_interval() passed
> sched_rr_get_interval01.c:64: TPASS: Time quantum 0s 99999990ns
> sched_rr_get_interval01.c:72: TFAIL: /proc/sys/kernel/sched_rr_timeslice_ms != 100 got 90
> sched_rr_get_interval01.c:57: TPASS: sched_rr_get_interval() passed
> sched_rr_get_interval01.c:64: TPASS: Time quantum 0s 99999990ns
> sched_rr_get_interval01.c:72: TFAIL: /proc/sys/kernel/sched_rr_timeslice_ms != 100 got 90
> 
> What this test does is to compare the return value from the
> sched_rr_get_interval() and the sched_rr_timeslice_ms sysctl file and
> fails if they do not match.
> 
> The prolem it found is the intial sysctl file value which was computed as:
> 
> static int sysctl_sched_rr_timeslice = (MSEC_PER_SEC / HZ) * RR_TIMESLICE;
> 
> which works fine as long as MSEC_PER_SEC is multiple of HZ, however it
> introduces 10% rounding error for CONFIG_HZ_300:
> 
> (MSEC_PER_SEC / HZ) * (100 * HZ / 1000)
> 
> (1000 / 300) * (100 * 300 / 1000)
> 
> 3 * 30 = 90
> 
> This can be easily fixed by reversing the order of the multiplication
> and division. After this fix we get:
> 
> (MSEC_PER_SEC * (100 * HZ / 1000)) / HZ
> 
> (1000 * (100 * 300 / 1000)) / 300
> 
> (1000 * 30) / 300 = 100
> 
> Signed-off-by: Cyril Hrubis <chrubis@suse.cz>
> CC: Jiri Bohac <jbohac@suse.cz>

Acked-by: Mel Gorman <mgorman@suse.de>

-- 
Mel Gorman
SUSE Labs

Re: [PATCH 2/2] sched/rt: sysctl_sched_rr_timeslice show default timeslice after reset

[Mel Gorman]

On Wed, Jul 19, 2023 at 12:37:43PM +0200, Cyril Hrubis wrote:
> The sched_rr_timeslice can be reset to default by writing value that is
> <= 0. However after reading from this file we always got the last value
> written, which is not useful at all.
> 
> $ echo -1 > /proc/sys/kernel/sched_rr_timeslice_ms
> $ cat /proc/sys/kernel/sched_rr_timeslice_ms
> -1
> 
> Fix this by setting the variable that holds the sysctl file value to the
> jiffies_to_msecs(RR_TIMESLICE) in case that <= 0 value was written.
> 
> Signed-off-by: Cyril Hrubis <chrubis@suse.cz>
> CC: Jiri Bohac <jbohac@suse.cz>

Acked-by: Mel Gorman <mgorman@suse.de>

-- 
Mel Gorman
SUSE Labs

Re: [PATCH 2/2] sched/rt: sysctl_sched_rr_timeslice show default timeslice after reset

[Petr Vorel]

Reviewed-by: Petr Vorel <pvorel@suse.cz>
Tested-by: Petr Vorel <pvorel@suse.cz>

Kind regards,
Petr

Re: [LTP] [PATCH 1/2] sched/rt: Fix sysctl_sched_rr_timeslice intial value

[Petr Vorel]

Tested-by: Petr Vorel <pvorel@suse.cz>

Kind regards,
Petr